AI & the Goldilocks Of Regulation
I will focus this article on previous examples of industry and technology regulation that can be used as a framing mechanism for our current situation with AI. This post will address what happens when Regulation is too rigidly applied, loosely interpreted, or too late to make a more significant impact.
Warning: This article includes some strange references to events that were footnotes in history books you read when you were 14.
Overboard Regulation
British Navigation Acts
It's a deep cut, for sure. Requiring all colonies to trade exclusively with Great Britain didn’t go well. Of course, other factors led to revolutions in various colonies, but the British Navigation Acts were too draconian. America wasn’t the only colony that rebelled, benefitting from the generosity of the empire’s enemies and frienemies. Smuggling increased significantly, as did corruption and payoffs to look the other way.
Regulation isn’t simple, and it's no surprise that applying simple, silver-bullet solutions doesn’t work.
The Prohibition Era
A 13-year nationwide ban on the production and sale of alcohol gave us many things. Cool stuff like speakeasies and hot rods were accompanied by an increase in organized crime and an overwhelmed law enforcement and judicial system. Of course, taxing a widespread vice and limiting its use to an arbitrary age has been far more lucrative than making it illegal. It reminds me of Alcohol’s BFF, Tobacco, which you can get legally at an even earlier age.
Drinking Age Act of 1984
This seemed like the next logical example. Stigmatizing alcohol consumption has only increased its popularity with those under 21 here in the US. The result has been an absolute shitshow (one I was guilty of), as binge drinking and blacking out is not widely seen as abusing a drug. If alcohol was introduced responsibly, it might have a better chance of being used responsibly.
Pretending that underage drinking doesn’t happen until something awful happens that we forget about until it happens again isn’t an effective policy. Applying this thinking has had dangerous consequences in different contexts.
Absence of Regulation
Lack of regulation? I could cite the following:
Unregulated Stock Market Practices and the Great Depression.
The deregulation of Savings and Loan Institutions that led to the S&L crisis of the 1980s.
Accounting Fraud and lack of oversight drove the Enron Scandal in the 2000s.
Subprime Mortgage Lending and lack of oversight in derivatives markets and The 2008 Global Financial Crisis.
The recent Cryptocurrency Scandals due to little/no regulation
The list of well-known and acknowledged financial regulation missteps goes on. The market can't just self-regulate; we keep giving it opportunities to learn from its mistakes just in case—but enough economic examples. Let’s look at some social and software stuff instead.
The Home Run Race of 1998
Steroids and ballplayers in the late 90s went together like today’s tech bros and podcasts. Everyone, whether or not they were into sports, knew McGuire and Sosa (and many others) were on steroids or some ungodly combination of chemicals. It was impossible to deny, as these men were…. straight-up yoked overnight. The MLB turned a blind eye to the issue because it sold tickets and continued to do so when, years later, Bonds broke the record again.
It’s challenging to acknowledge that the home run record was broken when the clear winner was Steroids.
Silk Road
I can’t tell you the first time I heard about Silk Road, but it sounded so crazy that I shrugged it off. Cryptocurrency was so new that I think I asked about the exchange rate with the Linden Dollar and started an argument by accident. When Silk Road was shut down, its competition only got stronger. Dark web marketplaces are now an international police game of whack-a-mole that could have been avoided if common-sense regulation had been implemented.
Kia Boys
Technology outpaced regulation in such an odd way that it's no wonder there are so many lawsuits over this instance. Kia decided to differentiate by offering premium digital features in a low-end automotive product that wasn't subject to enough QA. Social Media amplified software hacks, and teenagers did what they do best: dumb things like joyriding and destroying property. Kia will pay a price, but they seem plagued by repeating their security mistakes.
In the examples above, someone in an influential role decided that the impact of not acknowledging a problem was more manageable than addressing it. MLB was seeing a spike in revenue, Silk Road was a complicated issue to address, and Kia likely thought the lawsuits were worth the risk of letting competition reduce their market share. Mistakes are always clearest in the rearview mirror.
Hindsight Regulation
Sadly, it can take a tragic turn of events to shift sentiment enough to result in laws being written. The silver lining to these scenarios is that they are often so awful or widespread that enforcement is slightly less of a challenge–the situations listed below have had, or hopefully will have, long-lasting impacts.
Triangle Shirtwaist Factory Fire
Another Deep Cut, I know. One of the deadliest industrial disaster events in US history took place in a NYC sweatshop in 1911. 146 garment workers died in a building east of Washington Square Park due to unsafe working conditions, including but not limited to locked stairwells and exits and inadequate fire prevention. Modern workplace safety wouldn’t be the same without this incident, yet the conditions that led to this event weren’t uncommon and could have been avoided.
Cuyahoga River Fire
A classic anti-pattern. Cleveland, Ohio, wasn’t the only city with factories and towns that dumped industrial waste into rivers. Local waste disposal laws were on the books but weren’t well-enforced. After decades of environmental abuse, it took the Cuyahoga River to set itself on fire for 30 minutes for the Clean Water Act of 1970 to pass. The worst part? It wasn’t the first or most damaging time it had been on fire.
Asbestos
Enough about fires; let's talk about benign neglect. Asbestos was everywhere in construction and manufacturing, from automotive brakes to the ceiling tiles in a house I bought in Chicago. Manufacturers knew the health risks, which led to many preventable diseases and mesothelioma. OSHA drove legislation that resulted in its ban, but the damage had already been done. Health Insurance isn't cheap, and ignoring issues like Asbestos is one of many drivers.
Hinsight Regulation has a notable impact, but only after the right combination of press coverage, time, and societal pressure forces action. The challenge with emerging technology is that we can't estimate the impact because we need to know what it can or can't do.
Technology is a tool in the hands of humans, who choose how to wield it--until you build technology designed to act without guidance.
Goldilocks and the machines
If we don’t implement common-sense regulations for AI, we could end up in a Hindsight Regulation scenario, whereby the damage has already been done at scale, and we want to prevent it from happening again. In this case, we must learn a lesson the hard way after failing to do the right thing.
Should we fail to regulate an industry or technology or pretend like issues aren’t happening…the damage tends to increase to the degree that it impacts entire economic systems. Failing to act until it’s too late relies on the hope that lawsuits will be damaging enough to individuals/organizations/brands to prevent bad actors from attempting such activities.
Let’s face it: AI overregulation is not on the horizon.
We don’t have to worry about extreme measures and bootleg AI use on the fringes of society. I suppose the price we pay is the erosion of agreed-upon truths, lack of objective facts, and potentially massive economic disruption. Scenarios, suggestions, and my POV? That’s the final two installments in the series. Stick around; it's about to get good.